自行编译XDL，DIEN单机训练无法运行

x-deeplearning/xdl-algorithm-solution/DIEN/script/ 实例无法运行，提示错误如下：
# python train.py --run_mode=local --config=config.json
config: {u'checkpoint': {u'save_interval': 20000, u'output_dir': u'./ckpt_dir/'}, u'job_type': u'train', u'model': u'dien', u'seed': 3, u'data_dir': u'../data'}
Traceback (most recent call last):
File "train.py", line 127, in
train()
File "train.py", line 71, in train
train_ops = model.build_final_net(EMBEDDING_DIM, sample_io)
File "/root/x-deeplearning/xdl-algorithm-solution/DIEN/script/model.py", line 291, in build_final_net
*self.xdl_embedding(datas, EMBEDDING_DIM, *sample_io.get_n()))
File "/usr/local/lib/python2.7/dist-packages/xdl-1.0-py2.7.egg/xdl/python/backend/tf/tf_backend.py", line 160, in _wrapper
input = recursive_make_placeholder(x, xdl_inputs, placeholders)
File "/usr/local/lib/python2.7/dist-packages/xdl-1.0-py2.7.egg/xdl/python/backend/tf/tf_backend.py", line 49, in recursive_make_placeholder
placeholder = make_placeholder(x)
File "/usr/local/lib/python2.7/dist-packages/xdl-1.0-py2.7.egg/xdl/python/backend/tf/tf_backend.py", line 76, in make_placeholder
if len(x.shape) > 1:
TypeError: object of type 'NoneType' has no len()

Start by running `python train.py --run_mode=local --config=config.json` in `DIEN/script/` and inspect the calls from `train.py` into `model.py`, especially `build_final_net` and `xdl_embedding`. Then compare them with the placeholder handling shown in `xdl/python/backend/tf/tf_backend.py`; done means local DIEN training starts without the reported NoneType shape error.

Written by the indexing model from the issue text.